2-way: Waveguide + Cardioid-like

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
After dipoles it seems waveguides are the new craze. In spite of all the differences between the two, the concepts share a focus on controlling directivity – which was also the main criterion in the design of my Unbaffled Dipole. Therefore, for me it wasn’t much of a stretch to design a speaker with a waveguide.

The concept is a 2-way speaker with a 12” woofer and a 1" compression driver with a 12” waveguide, crossed passively at 1 khz. A miniDSP is used for amplitude corrections and the crossover to subwoofers below about 100 hz.

I considered many drivers, but I ultimately settled for a Selenium D220Ti compression driver, a Dayton H12RW Waveguide and a A&D R1230 midwoofer. The Seleniums are relatively cheap and they are quite popular, so I thought those would be a good starting point. I bought both the Dayton JBL-clone (the one used in the Econowaves) an the Dayton H12RW. I chose to use the H12RW because it is axi-symmetric and thus has more constant vertical control. The rectangular waveguide allows better vertical alignment of the acoustic centres, which helps to minimize lobing at the crossover, but I don’t believe that to be as relevant as some others do. With a crossover at 1 khz, with this speaker the acoustic centres are spaced less than one wavelength apart, which is much better than many a conventional 2-way box speaker.

Initially I intended to put the drivers in a closed box, but in order to get the smoothest possible powerresponse through the crossover-region a simple box seemed not to be the best solution. Although others have already shown that the closed box format is very workable if you cross at the frequency where the directivity of the woofer and waveguide match, below the crossover frequency directivity usually decreases quite fast. Therefore I decided to experiment with an open box – a box with holes in it. The box is stuffed to the brim with rockwool (I intended to use glasswool, my favorite, but I didn’t have it at the time) and I use carpet without backing to finish the sides.

Some pictures:
611117d1491990102-2-way-waveguide-cardioid-like-1.jpg

611118d1491990102-2-way-waveguide-cardioid-like-2.jpg

611119d1491990102-2-way-waveguide-cardioid-like-3.jpg

611120d1491990102-2-way-waveguide-cardioid-like-4.jpg

611121d1491990102-2-way-waveguide-cardioid-like-5.jpg


I think it is difficult to predict the exact acoustic behavior of this concept. My expectation was that the directivity pattern would be more or less dipolar at low frequencies and that it would lie somewhere between that of a conventional box a cardioid at the top of the intended passband. Such behavior would be more or less ideal, as this would help to match DI’s at the crossover and it would not let directivity drop too much at lower frequencies.

Typical measurement setup (measurements of the speaker with crossover are taken about 1.5 meters higher than this) :

611122d1491990102-2-way-waveguide-cardioid-like-6.jpg


Let’s see some measurements!

611123d1491990102-2-way-waveguide-cardioid-like-ad-woofer-behuizing-vol-met-demping-0-tot-90.jpg
A&D R1230 Horizontal dispersion from 0 to 90 degrees in 15 degree steps (gated measurement).



These measurements show that practice approximates the anticipated behavior pretty well. All curves look pretty smooth and there is good directivity around 1 khz. Below about 300 hz the radiation pattern looks like a dipole, above that frequency there is much greater directivity – in the transition region some widening is apparent. What happens below 100 hz is unclear.

Now what does it look like at the rear?

611137d1491992191-2-way-waveguide-cardioid-like-ad-woofer-behuizing-vol-met-demping-2c-achterzijde.jpg
A&D R1230 Horizontal dispersion from 180 to 90 degrees in 15 degree steps (gated measurement).

It looks a bit messy, but at least all curves are a lot lower in level than those at the front. Interesting to observe is that the 135 degree curve overall looks to be lowest in level, which is a bit like a super-cardioid. When you look at the non-weighted average of all these curves you get a pretty smooth looking curve.

[not available]

In the next picture you see no new data, but it's still informative: A&D R1230 comparison of 0 degree and 180 degree response:

611143d1491992929-2-way-waveguide-cardioid-like-ad-woofer-behuizing-vol-met-demping-vergelijk-voor-en-achter.jpg
A&D R1230 Comparison response at 180 and 0 degrees (gated measurement).

At low frequencies the difference between the curves is greater than it would be with a closed box, at higher frequencies it’s less.

Let’s have a look at what the Selenium-Dayton combo looks like in comparison!

611124d1491990102-2-way-waveguide-cardioid-like-selenium_d220ti-en-ad_r1230-onaxis.jpg
Comparison on-axis response A&D R1230 and Selenium D220Ti on Dayton HR12W

The first thing I noted was the ragged response above 5 khz. Let’s hope it doesn’t sound as ugly as it looks! The second thing that struck me as odd is that except for the 2khz bump the curve overall is very flat! I expected to see a downward trend. Let’s see how constant directivity is.

Selenium D220Ti on Dayton HR12W Horizontal dispersion from 0 to 90 degrees in 15 degree steps. Measurements taken at approx. 2 meters.

An externally hosted image should be here but it was not working when we last tested it.
Selenium/Dayton Horizontal dispersion from 0 to 90 degrees in 15 degree steps (gated measurement).

In comparison to my dipoles, this is not very constant directivity, however it does look very controlled! When you look at greater off-axis angles you see increasingly downward sloping curves. It doesn’t look bad, but making this flat on-axis might lead to a sound that is not very bright. The Daytons are axi-symmetric. Therefore I expected to see a diffraction dip in the treble region, but strange enough there appears to be nothing there! Ignoring the increasing downward slope, the off-axis curves track the on-axis curve very well. I would say this is exactly what you want, but not everyone agrees: http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-418.html#post2146754. I wonder if HOM are the ‘cause’ of this slightly unexpected yet welcome behavior. Too bad I don’t know how to measure HOM.

The passive crossover was a bit more difficult to get right than I’m used to. The acoustic center of the tweeter lies a bit behind that of the woofer. This causes some excess phase. To work around this, I had to cross steep and I had to do a few tricks to get the right amount of relative phase rotation [EDIT: as speaker dave later remarked, there is still some excess phase left in both the measurement on the tweeter and the midwoofer, which makes the plots look a bit messy]

611125d1491990102-2-way-waveguide-cardioid-like-filter.jpg
Crossover schematic

Some inductors have unnecessary low Rdc, but that’s because I used components I already had. When I build the crossover into the speaker I’ll use more logical component values. The impedance correction is optional.

611126d1491990102-2-way-waveguide-cardioid-like-overdrachtsfunctie.jpg
Transfer function


611131d1491990336-2-way-waveguide-cardioid-like-response-15-graden.jpg
Response at 15 degrees


611132d1491990336-2-way-waveguide-cardioid-like-fase-15-graden.jpg
Phase of both drivers including crossover, at 15 degrees

EQ is done with a miniDSP. After some tweaking I ended up with this result:

611133d1491990336-2-way-waveguide-cardioid-like-afstraalgedrag-0-tot-90-gated-2c-1-.-24-oct.jpg
2-Way dispersion 0 to 90 degrees (gated measurement) EDIT: taken at a later date and more flat than the original response.

[not available] [2-Way comparison between on-axis response and the non-weighted average response from 0 degrees to 180 degrees [/I]

Unfortunately, then it started to rain, so I couldn’t do more measurements. Indoors I later lifted the highs a little to make the response flat on the listening axis of about 20 degrees. Also I EQ’ed a few of those nasty looking peaks. In the range below 500 hz I do a little room-EQ. Subwoofers are used below about 100 hz.

The design is not 100% ready yet. I’ll probably keep tweaking the EQ for some time. I’ve noticed that EQ is very critical with this speaker. When it’s ready, I may do a short subjective review. Thus far I am very pleased with the sound. All in all I think it is better than my dipoles. It all just sounds so effortless and very detailed. I tend to listen to them at a much higher sound level than other speakers. At normal sound levels they sound very clean and clear. At higher levels the midrange becomes a bit shouty on some material.

I’d like to hear what you guys think!

Some observations that I find interesting and which might be worth a discussion:

- The ugly looking high frequency response doesn't sound half as ugly as it looks. This may have to do with the critical bandwidth of the auditory system.
- The unfiltered response of the CD-waveguide combo doesn't have the expected downward trend. A downward trend corresponds with more or less constant directivity. Personally I wouldn't call the directivity of this combo constant (at least not as constant as is attainable with dipoles with minimum baffle), but it's not really less constant than that of other speakers that their designers do call Constant Directivity. Why is the on-axis response relatively flat?
- There is no on-axis dip anywhere. Does this mean this combination creates a lot of HOM? If so:
- Don't HOM primarily harm the highest frequencies? If so, then I don't understand, because the highs sound very clean. If there is one thing I don't like about them, it is perhaps that they sound a bit less spacious than the highs of wider dispersion speakers.
- At higher levels the midrange becomes a bit shouty on some material. Do you guys things this is caused by some kind of horn distortion, or maybe be the rather lively room they are in? Another possibility is the EQ that may not yet be top notch.
 
Last edited:
Can you explain some more about your intention of having carpet over the holes? A very "lossy box" perhaps? I'm experimenting with fish net material behind a solid baffle, so the sides of the "box" are nothing more than glass wool suspended behind the drivers. Automotive air filters are another way of attacking this as they create very lossy sides, close around the driver. The midrange is phenomenal.
 
Great job indeed! :D

...

- The ugly looking high frequency response doesn't sound half as ugly as it looks. This may have to do with the critical bandwidth of the auditory system.
....

As I saw the throat of the WG, there's still that old problem. Have you seen this:

http://www.diyaudio.com/forums/mult...baffle-eminence-skytronic-35.html#post2340183

It's not entirely the same situation as yours. But the throat is worth investigating and some improvement.

The entrance of this Dayton WG is not exactly 1", it's slightly bigger. So there'd be a step from the driver to the WG.
 
Last edited:
Very nice!

That 150-500 Hz increase in pressure is likely all baffle related.

Look at MBK's experiments (and the difference in baffle/pressure release) here:

http://www.diyaudio.com/forums/multi-way/142691-adventures-cardioid.html

Thanks. I've seen that topic before. Do you mean that somewhere in that topic 'baffle related' issues are discussed? I browsed the thread again but I couldn't find anything directly related.

Fantastic Keyser! Excellent job. Thank you for posting and having the insight to do this.

Dan

My pleasure. Thanks for the kind words.


Very Nice!!!! :D:D:D

Look up aperiodic enclosures - it is similar to the method you are using on your enclosure and might help out with the crossover.

There are several websites that discuss the theory - here is one that shows a practical application: WD25A Design Pg2


There are some similarities (like the free-air resonance of the woofer in my speaker is 58 hz, but when built into the enclosure it drops to 52 hz - and the impedance peak at resonance drops from about 40 ohms to about 20), but the design criteria are very different. The aperiodic enclosure is designed for bass, but I designed my speaker with primarily directivity in mind.

Great job indeed! :D



As I saw the throat of the WG, there's still that old problem. Have you seen this:

http://www.diyaudio.com/forums/mult...baffle-eminence-skytronic-35.html#post2340183

It's not entirely the same situation as yours. But the throat is worth investigating and some improvement.

The entrance of this Dayton WG is not exactly 1", it's slightly bigger. So there'd be a step from the driver to the WG.

You're right, the waveguide and the compression driver don't match perfectly. Do you think this is what causes the ripples in the frequency response? The irregularities are the same at every angle, therefore I reckoned it was the Selenium.


By the way, yesterday I set the entire system up in the garden. First I undid some roomEQ in the midrange and I made the response very flat. Then I decided to have a listen. It seems the harshness in the midrange seems to be due to faults in the frequency response. Outside it sounded very clean, also at higher levels. The biggest issue was the floor-bounce, which degrades the frequency response much more than it does indoors, where its effect is mitigated by other reflections.

After yesterdays corrections the speaker sounds a lot better. I've decided to optimize the speaker placement based on measurements taken at the listening position and not to do any roomEQ on the speakers anymore. The subs still do have some EQ. This sounds very good to my ears.

To me this simple plastic waveguide and inexpensive compression driver sound no less sweet than the best cone-dome combinations. I'm not sure if I'm supposed to hear horn honk or HOMs maybe. What do those of you with more experience with horns think?
 
Thanks. I've seen that topic before. Do you mean that somewhere in that topic 'baffle related' issues are discussed? I browsed the thread again but I couldn't find anything directly related.

http://www.diyaudio.com/forums/multi-way/142691-adventures-cardioid.html

You only need to read posts 1, 2, 4, 5, 6, 7, 8, and 10 in context with the measurements and the pictures. (post 3 is exclusively concerning the horn and tweeter.)

1 & 2 is an open baffle with a decidedly non-dipole polar pattern in the midrange - because of the overall profile of the baffle.


4,5, 6, and 7 all make use of a larger baffle (though with the "muffler", and as a result the pattern becomes "wider" or less directive than seen in post 8. (..graph in post 7 vs. post 8.) The baffle looks smaller than yours - so the polars become more directive higher in freq.. (..in this case starting at about 250 Hz and up).



Post 8 starts off with a "title in bold" of:

"So let's eliminate the baffle"

It's there that you get to see where the directivity is fairly uniform - because he removed the baffle. (..or actually there is still the small mounting "ring" baffle for driver and muffler.)

You can see a picture of it resting on top of the old baffle in post 10.



I suppose this is all a long-winded way of suggesting that you try "spliting" the baffle up from the horn/CD and the midbass driver. (..two separate "Baffles" stacked but with a modest amount of air-space between them.)
 
Last edited:
Scott, thanks. I've looked at it some more, but it still isn't completely clear to me. Maybe someday I'll experiment some more with it to get a better grasp on what's really happening in the transition region between dipole with damped rear-wave and the more cardioid/closed-box like behavior at higher frequencies.

I was under the assumption that last week I uploaded these measurement graphs, but it appears I only uploaded the sonogram.

The measurements in the opening post show an excess of energy below about 500 hz and recessed high frequencies. Those issues have been dealt with, as well as some other small irregularities. I'm also showing the more smoothed, ungated responses to better show what's going on at lower frequencies. The final curve is the average response at the rear.
 

Attachments

  • afstraalgedrag 0 tot 90 gated, 1 . 24 oct.jpg
    afstraalgedrag 0 tot 90 gated, 1 . 24 oct.jpg
    125.5 KB · Views: 4,569
  • afstraalgedrag 0 tot 90 ungated, 1 . 2 oct.jpg
    afstraalgedrag 0 tot 90 ungated, 1 . 2 oct.jpg
    118.8 KB · Views: 2,539
  • Unweighted average response rear, curves at 180 deg. to 90 deg. in 15 deg. steps..jpg
    Unweighted average response rear, curves at 180 deg. to 90 deg. in 15 deg. steps..jpg
    89.9 KB · Views: 2,107
Cool project...

The first thing I noted was the ragged response above 5 khz. Let’s hope it doesn’t sound as ugly as it looks!

IMO, the Selenium to me is the worst sounding CD of them all, it could be just the screw on types though ( I never did get anything I liked out of the Dayton + Selenium CD). In that price range I would choose the Celestion 1745 over it.
 
Last edited:
:D

Thanks for sharing.

The opinions of the Selenium vary almost as widely as its amplitude response above 6 khz! I think if you look at the response curve (or the CSD, for that matter) long enough, you just have to hear lousy high frequencies!

However, it is interesting that of the several audiophiles I've had over to listen to these speakers - most of them with limited technical knowledge of loudspeakers - nobody commented about the highs being sharp or lacking resolution! I myself don't hear it either.

I would like to know what others think of it too, though!
 
No problem, could be true about staring at measurements and then 'hearing' it but I listened to 5 CDs I bought for my 2-way waveguide build before I measured any of them.

There are many that have enjoyed the Selenium driver in the econo waveguides. Im just not one of them.
 
Last edited:
The passive crossover was a bit more difficult to get right than I’m used to. The acoustic centre of the tweeter lies a bit behind that of the woofer. This causes some excess phase. To work around this, I had to cross steep and I had to do a few tricks to get the right amount of relative phase rotation.

There was a good discussion about this on HTguide.com when Augerpro was building a two. I will try and find it. We also rear mounted our woofers to try and get the AC closer together.
 
I think this was the discussion

John Kreskovsky brought up an interesting alignment that might be prefect to try with this design, for the passive version. Here is what he said:

"Additionally, a 3rd order, quasi Linkwitz type crossover is easily constructed by cascading a B2 with a B1 filter and delaying the high pass section by 0.25 wave lengths at the crossover frequency. The response error will be +0, -1.2dB max. This is easily constructed using passive elements.

For the HP,

C1 = 1/(4 Pi R F), C2 = 1.2 /(Pi R F), L = 1/(2.4 Pi R F)

For the LP

L1 = R/(Pi F), L2 = R/(4.8 Pi F), C = 0.6 /(Pi R F)

Offset tweeter by 0.25 wave lengths at the x-o frequency and connect with inverted phase. "

What caught my eye was that the AC offset (versus 1300hz crossover) of my No Quarter design is nearly identical to the above spec of .25 wavelength of crossover. I also like that this has a broader overlap than LR4, but won't be as demanding as LR2. And all the usual LR properties of symmetric lobing and in phase drivers.

Found here
HTGuide Forum - Can you smell what Brando's cooking!?
 
D220Ti crossover point

I'm also using the D220Ti + HW12 in a speaker that I am building so I read your thread with much interest.

I noticed that you cross over at 1k Hz. This is on the low side for the compression driver. I am wondering if you have done any listening or distortion tests yet, and if so what is going on around 1k Hz? Can you detect any strain, etc? Maybe this is why the midrange sounds "shouty" (per your initial post)?

Nice work, BTW!

-Charlie
 
Last edited:
I'm also using the D220Ti + HW12 in a speaker that I am building so I read your thread with much interest.

I noticed that you cross over at 1k Hz. This is on the low side for the compression driver. I am wondering if you have done any listening or distortion tests yet, and if so what is going on around 1k Hz? Can you detect any strain, etc?


Hi Charlie,

I haven't done distortion measurements yet (will do that during next measurement session), but I'm convinced the Selenium is used nowhere near its limits. If these speakers were to be used in a PA, I would probably have used a crossover closer to 1.5 khz, but for hifi this is not necessary. Horn-loading is still evident around 700 hz, under which frequency the response begins to roll off. The crossover at 1 khz is about 30 dB acoustically, which is significant, so I don't see any problem here.

Last saturday I threw a party to celebrate my birthday. Some friends were interested in my new speakers and we listened to Child in Time by Deep Purple, from their Made in Japan album. Right now the limiting factor is probably the power amp (but it might also be the output of the miniDSP), an NAD C370 integrated amp with some 150 watts at 8 ohms. Up to about 107 dB continous inroom SPL it remained very clean, but above that it began to clip. This is probably due to the significant boost at 100 hz (about 8 dB's), to compensate for the open-box roll-off. I'm okay with a 107 dB SPL maximum continuous level - those aren't levels I want to endure too often or too long, let alone even higher levels.

Maybe this is why the midrange sounds "shouty" (per your initial post)?

Nice work, BTW!

-Charlie

The slightly shouty character I mentioned disappeared after I omitted the roomEQ in the midrange, improved the EQ of the speaker and changed the placement of the speakers a bit. Now they sound very good to me. I was really happy with the sound of the dipoles I had before (link in the opening post), but I think these sound even better.

EDIT: Do have topic about your own speakers. I saw a picture in the 3-way econowave thread. Now I'd like to see some more! Also I'd like to hear about your design choices.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.